U.S. patent application number 12/105449 was filed with the patent office on 2008-08-14 for structure for an asynchronous data interface.
Invention is credited to Scott J. Lemke, Kevin N. Magill, Michael S. Siegel.
Application Number | 20080195774 12/105449 |
Document ID | / |
Family ID | 39686828 |
Filed Date | 2008-08-14 |
United States Patent
Application |
20080195774 |
Kind Code |
A1 |
Lemke; Scott J. ; et
al. |
August 14, 2008 |
STRUCTURE FOR AN ASYNCHRONOUS DATA INTERFACE
Abstract
A design structure embodied in a machine readable storage medium
for designing, manufacturing, and/or testing a design, the design
structure comprising for an interface system is disclosed. The
system includes a buffer that receives data from a source in a
first clock domain and stores the data to be read by a destination
in a second clock domain, wherein the buffer functions in both the
first clock domain and the second clock domain; a write pointer
that points to data written by the source; and a read pointer that
points to data read by the destination. According to the design
structure, the write pointer and the read pointer are utilized to
enable the data to be transmitted from the first clock domain to
the second clock domain asynchronously.
Inventors: |
Lemke; Scott J.; (Raleigh,
NC) ; Magill; Kevin N.; (Raleigh, NC) ;
Siegel; Michael S.; (Raleigh, NC) |
Correspondence
Address: |
IBM CORPORATION, INTELLECTUAL PROPERTY LAW;DEPT 917, BLDG. 006-1
3605 HIGHWAY 52 NORTH
ROCHESTER
MN
55901-7829
US
|
Family ID: |
39686828 |
Appl. No.: |
12/105449 |
Filed: |
April 18, 2008 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
11622445 |
Jan 11, 2007 |
|
|
|
12105449 |
|
|
|
|
Current U.S.
Class: |
710/57 |
Current CPC
Class: |
G06F 13/4059
20130101 |
Class at
Publication: |
710/57 |
International
Class: |
G06F 13/00 20060101
G06F013/00 |
Claims
1. A design structure embodied in a machine readable storage medium
for at least one of designing, manufacturing, and testing a design,
the design structure comprising: an interface system comprising: a
buffer that receives data from a source in a first clock domain and
stores the data to be read by a destination in a second clock
domain, and wherein the buffer functions in both the first clock
domain and the second clock domain; a write pointer that points to
data written by the source; and a read pointer that points to data
read by the destination, wherein the write pointer and the read
pointer are utilized to enable the data to be transmitted from the
first clock domain to the second clock domain asynchronously.
2. The design structure of claim 1 wherein the data can be
continuously transmitted from the source to the destination as long
as the buffer is neither full nor empty.
3. The design structure of claim 2 further comprising: a first
determination unit that determines whether the buffer is full; and
a second determination unit that determines whether the buffer is
empty.
4. The design structure of claim 2 further comprising: a first
determination unit that determines whether the buffer is full by
comparing the write pointer to a copy of the read pointer; and a
second determination unit that determines whether the buffer is
empty by comparing the read pointer to a copy of the write
pointer.
5. The design structure of claim 1 wherein a copy of the write
pointer is transmitted from the first clock domain to the second
clock domain.
6. The design structure of claim 1 wherein a copy of the read
pointer is transmitted from the second clock domain to the first
clock domain.
7. The design structure of claim 1 further comprising a hold unit
that delays an acknowledgment that either the copy of the write
pointer or the copy of the read pointer has been received.
8. The design structure of claim 1, wherein the design structure
comprises a netlist, which describes the interface system.
9. The design structure of claim 1, wherein the design structure
resides on the machine readable storage medium as a data format
used for the exchange of layout data of integrated circuits.
10. A design structure embodied in a machine readable storage
medium for at least one of designing, manufacturing, and testing a
design, the design structure comprising: a system comprising: a
processor; a buffer coupled to the processor, wherein the buffer
receives data from a source in a first clock domain and stores the
data to be read by a destination in a second clock domain, wherein
the buffer functions in both the first clock domain and the second
clock domain; a write pointer that points to data written by the
source; and a read pointer that points to data read by the
destination, wherein the write pointer and the read pointer are
utilized to enable the data to be transmitted from the first clock
domain to the second clock domain asynchronously.
11. The design structure of claim 10 wherein the data can be
continuously transmitted from the source to the destination as long
as the buffer is neither full nor empty.
12. The design structure of claim 11 further comprising: a first
determination unit that determines whether the buffer is full; and
a second determination unit that determines whether the buffer is
empty.
13. The design structure of claim 11 further comprising: a first
determination unit that determines whether the buffer is full by
comparing the write pointer to a copy of the read pointer; and a
second determination unit that determines whether the buffer is
empty by comparing the read pointer to a copy of the write
pointer.
14. The design structure of claim 10 wherein a copy of the write
pointer is transmitted from the first clock domain to the second
clock domain.
15. The design structure of claim 10 wherein a copy of the read
pointer is transmitted from the second clock domain to the first
clock domain.
16. The design structure of claim 10 further comprising a hold unit
that delays an acknowledgment that either the copy of the write
pointer or the copy of the read pointer has been received.
17. The design structure of claim 8, wherein the design structure
comprises a netlist, which describes the system.
18. The design structure of claim 8, wherein the design structure
resides on the machine readable storage medium as a data format
used for the exchange of layout data of integrated circuits.
Description
CROSS-REFERENCE TO RELATED APPLICATION
[0001] This application is a continuation-in-part of co-pending
U.S. patent application Ser. No. 11/622,445, filed Jan. 11, 2007,
which is herein incorporated by reference.
BACKGROUND OF THE INVENTION
[0002] 1. Field of Invention
[0003] The present invention generally relates to design
structures, and more specifically, design structures for computer
systems, and more particularly to an asynchronous data
interface.
[0004] Asynchronous data interfaces are well known. An asynchronous
data interface generally connects two systems such as a processor
core and an external bus, where the two systems operate in
independent time domains. In other words, there are two different
clock domains, each domain having its own clock with different
rates, frequencies, phases, etc.
[0005] In order to transmit data from one clock domain to another
clock domain, the data typically needs to be synchronized between
the two clock domains. One solution is to rely on a fixed frequency
difference between the two sides of the interface to transfer a
packet of data. A problem with this solution is that the
frequencies need to be known in order to determine when data may be
transmitted from one system to another. For the frequencies to be
known, the frequencies need to be calculated, which requires
additional system time and resources.
[0006] Another solution is utilizing data interfaces that use a
single bit when crossing clock domains to avoid multiple bits being
latched in different clock cycles. Once the single bit safely
reaches the destination side, the full data width may be sampled.
However, the single bit is not considered "safe" until
metastability is resolved. Metastability is usually resolved by
latching the bit two or more times in the destination clock domain.
Once the full data width is sampled by the destination side, it
must signal to the source side that data has been accepted. This is
done by sending another single bit acknowledgement to the source
side. In other words, the source side does not count a certain
number of clock cycles before the acknowledgement is seen. Once the
acknowledge bit safely reaches the source side, the source may drop
the data and begin the process again with new data. A problem with
this solution is that while it may work independently of clock
ratios, it limits the writing of subsequent data until the current
data has been read by the destination, which takes a certain number
of clock cycles (e.g., 6 clock cycles) to complete.
[0007] Accordingly, what is needed is an improved data interface.
The present invention addresses such a need.
SUMMARY OF THE INVENTION
[0008] A data interface system is disclosed. In one embodiment, the
system includes a buffer that receives data from a source in a
first clock domain and stores the data to be read by a destination
in a second clock domain, where the buffer functions in both the
first clock domain and the second clock domain; a write pointer
that points to data written by the source; and a read pointer that
points to data read by the destination.
[0009] According to the system and method disclosed herein, the
write pointer and the read pointer are utilized to enable the data
to be transmitted from the first clock domain to the second clock
domain asynchronously.
BRIEF DESCRIPTION OF THE DRAWINGS
[0010] FIG. 1 is a block diagram of an asynchronous data interface
system in accordance with one embodiment.
[0011] FIG. 2 is a flow chart showing a method for transferring
data from one clock domain to another clock domain in accordance
with one embodiment of the present invention.
[0012] FIG. 3 is a block diagram of the synchronization unit, the
write pointer unit, and the write sample unit, and the write copy
unit of FIG. 1, in accordance with one embodiment.
[0013] FIG. 4 is a block diagram of the advance hold logic unit of
FIG. 3 in accordance with one embodiment.
[0014] FIG. 5 is a block diagram of an advance hold logic unit,
which may be used to implement the advance hold logic unit of FIG.
3, in accordance with another embodiment.
[0015] FIG. 6 is a flow diagram of a design process used in
semiconductor design, manufacture, and/or test.
DETAILED DESCRIPTION OF THE INVENTION
[0016] The present invention relates to computer systems, and more
particularly to an asynchronous data interface. The following
description is presented to enable one of ordinary skill in the art
to make and use the invention, and is provided in the context of a
patent application and its requirements. Various modifications to
the preferred embodiment and the generic principles and features
described herein will be readily apparent to those skilled in the
art. Thus, the present invention is not intended to be limited to
the embodiments shown, but is to be accorded the widest scope
consistent with the principles and features described herein.
[0017] An data interface system is disclosed. In one embodiment,
the system includes a buffer that receives data from a source in a
source clock domain and stores the data to be read by a destination
in a destination clock domain. The buffer functions in both the
source clock domain and the destination clock domain. The system
also includes a write pointer that points to data written by the
source and includes a read pointer that points to data read by the
destination. The system transfers the data from the source clock
domain to the destination clock domain continuously as long as the
buffer is neither full nor empty. The system utilizes the write
pointer and the read pointer to determine whether the buffer is
full or empty. According to the system and method disclosed herein,
the write pointer and the read pointer enable the data to be
transmitted from the first clock domain to the second clock domain
asynchronously.
[0018] In one embodiment, a design structure embodied in a machine
readable storage medium for at least one of designing,
manufacturing, and testing a design is provided. The design
structure generally includes an interface system, which includes a
buffer that receives data from a source in a first clock domain and
stores the data to be read by a destination in a second clock
domain, and wherein the buffer functions in both the first clock
domain and the second clock domain. The interface system also
includes a write pointer that points to data written by the source,
and a read pointer that points to data read by the destination,
wherein the write pointer and the read pointer are utilized to
enable the data to be transmitted from the first clock domain to
the second clock domain asynchronously.
[0019] A design structure embodied in a machine readable storage
medium for at least one of designing, manufacturing, and testing a
design is provided. The design structure generally includes a
system that includes a processor, a buffer coupled to the
processor, wherein the buffer receives data from a source in a
first clock domain and stores the data to be read by a destination
in a second clock domain, wherein the buffer functions in both the
first clock domain and the second clock domain, a write pointer
that points to data written by the source, and a read pointer that
points to data read by the destination, wherein the write pointer
and the read pointer are utilized to enable the data to be
transmitted from the first clock domain to the second clock domain
asynchronously.
[0020] Although the present invention disclosed herein is described
in the context of write pointers, the present invention may apply
to read pointers, and still remain within the spirit and scope of
the present invention.
[0021] FIG. 1 is a block diagram of an asynchronous data interface
system 100 in accordance with one embodiment. The asynchronous data
interface system 100 includes a buffer 102, also referred to as a
rolling buffer. The buffer 102 includes a write port 104 in one
time domain (e.g., the source clock domain) for writing data into
the buffer 102, and includes a read port 108 in the other time
domain (e.g., the destination clock domain) for reading from the
buffer 102. In one embodiment, the write port 104 is controlled by
the source side, and the read port 108 is controlled by the
destination side. The asynchronous data interface system 100 also
includes a write pointer unit 112 for storing a write pointer, a
write sample unit 114 for storing a sampled version of the write
pointer to be sent to the destination side of the interface, a read
copy unit 116 for storing a sampled copy of the destination read
pointer, and a full determination unit 118 for determining when the
buffer 102 is full, and a synchronization unit 120. Any appropriate
types of write and read pointers may be used. The asynchronous data
interface system 100 also includes a read pointer unit 122, a read
sample unit 124 for storing a sampled version of the read pointer
for the source, a write copy unit 126 for storing a sampled copy of
the source write pointer, an empty determination unit 128, and a
synchronization unit 130. In one embodiment, the full and empty
determination units 118 and 128 may be implemented with
comparators, depending on the specific pointer implementation.
[0022] FIG. 2 is a flow chart showing a method for transferring
data from one clock domain to another clock domain in accordance
with one embodiment of the present invention. Referring to both
FIGS. 1 and 2 together, the process begins in step 202 where the
buffer 102 receives data from the source in the source clock
domain. Next, in step 204, the buffer 102 stores the data to be
read by the destination in a second clock domain. As described
above, the buffer functions in both the source clock domain and the
destination clock domain, via the write port being controlled by
the write pointer on the source side and the read port being
controlled by the read pointer on the destination side. As
described in more detail below, data can be continuously
transmitted from the source to the destination as long as the
buffer is neither full nor empty. Next, in step 206, the
asynchronous data interface system 100 utilizes the write pointer
and the read pointer to determine whether the buffer is full or
empty. As described in more detail below, the buffer 102, being in
both time domains, enables the source side to write to the buffer
102 in its own clock domain, and enables the destination side to
read from the buffer 102 in its own domain.
[0023] In one embodiment, the source updates the sampled write
pointer whenever it is not equal to the current write pointer, and
signals the destination that the source's sampled write pointer has
been updated. The source will then wait for an acknowledgement from
the destination that the sampled write pointer was seen before the
next update of the write pointer can occur. The source, via the
full determination unit 118, determines if the buffer 102 is full.
In one embodiment, to determine whether the buffer 102 is full, the
full determination unit 118 compares the current write pointer to a
copy of the read pointer received from the read copy unit 116. The
read copy unit 116 receives the copy of the read pointer from the
read sample unit 124, which asynchronously samples copies of the
read pointer.
[0024] In one embodiment, the write pointer unit 112, write sample
unit 114, write copy unit 126, read pointer unit 122, read sample
unit 124, and read copy unit 116 are initialized to a value of "all
0's" at system power on. Additionally, each unit is sized one bit
wider than is traditionally needed to index a buffer of a given
size (e.g., a buffer with 16 entries would use 5-bit pointers,
rather than 4-bit). The most significant bit (MSB) is considered a
"wrap bit" and is not used to access the buffer. This allows the
write and read pointers to "wrap" back to the first entry of the
buffer after writing/reading the last entry. The MSBs allow the
full and empty units to determine if one pointer has "wrapped"
relative to the other pointer and differentiate between an empty
condition and a full condition. Each time a data value is written
to the buffer, the write pointer is incremented by 1 and each time
a data value is read from the buffer, the read pointer is
incremented by 1. The buffer is empty when the write pointer
matches the read pointer, including the MSB. Note the system is
initialized to this state at power on. The buffer is full when all
bits of the write pointer, except the MSB, are equal to all bits of
the read pointer, except the MSB, and the MSB of the write pointer
is not equal to the MSB of the read pointer.
[0025] A benefit of the embodiments described herein is that their
implementations are not restricted to a particular brand of
"pointer math." For example, some conventional solutions require
the use of "Gray-coded pointers" in order to work properly, wherein
the embodiments described herein do not. The implementation of the
write pointer, read pointer, full detection, and empty detection
are left to the user. In one embodiment, the empty and full units
may be designed to handle the math of the read and write pointers
the same as a buffer implementation that existed in a single clock
domain. In other words, any method of tracking reads and writes in
a buffer may be chosen, including how to manipulate and detect the
pointers, and drop them into the system. With any method chosen,
the read and write pointers may be asynchronously copied between
clock domains so that empty/full units may perform read/write
pointer math.
[0026] In one embodiment, because the full determination unit 118
is comparing a current write pointer to a copy of the read pointer
which may be stale, the full determination unit 118 will be biased
to determine the buffer 102 as being full. Similarly, because the
empty determination unit 128 is comparing a current read pointer to
a copy of the write pointer which may be stale, the empty
determination unit 128 will be biased to determine the buffer 102
as being empty. This prevents over writing of data in a full buffer
and re-reading data in an empty buffer.
[0027] While the buffer 102 is not full, the source may continue to
write data to the buffer 102 and increment the write pointer until
the buffer 102 is full. In one embodiment, the buffer 102 is sized
such that in steady state operation the slower side of the
asynchronous data interface system is not throttled by the speed of
the interface.
[0028] In one embodiment, the destination will update the sampled
read pointer when the current read pointer does not equal the
sampled read pointer and signals the source that the sampled read
pointer has been updated. The destination will then wait for an
acknowledgement from the source that the sampled read pointer was
seen and copied before the next update of the sampled read pointer
can occur. The destination, via the empty determination unit 128,
determines if the buffer 102 is empty. In one embodiment, to
determine whether the buffer 102 is empty, the empty determination
unit 128 compares the current read pointer to a copy of the write
pointer received from the write copy unit 126. The write copy unit
126 receives the copy of the write pointer from the write sample
unit 114, which asynchronously samples copies of the write pointer.
While the buffer is not empty, the destination may read the data
from the buffer 102 and increment the read pointer until the buffer
102 has been emptied. If the destination is slower than the source,
the next update of the sampled write pointer will occur before the
buffer 102 is completely emptied, preventing an underflow condition
from occurring and optimizing the bandwidth of the interface.
[0029] Another benefit of the embodiments described herein is that
they work independently of clock ratios and they allow multiple
writes or reads without waiting for the other side to read/write.
Also, as described above, as long as the buffer is neither full nor
empty, data may be written to and read from the buffer 102 at an
efficient rate (e.g., a rate of 1 entry per clock cycle). Also, the
embodiments described herein sustain maximum throughput during
steady-state data movement and enable data movement across
asynchronous clock domains without relying on fixed frequency or
phase relationships between the domains. Accordingly, the interface
system works properly under any frequency and phase differences,
and under dynamically changing frequency and phase differences.
[0030] FIG. 3 is a block diagram of the synchronization unit 120,
the write pointer unit 112, and the write sample unit 114, and the
write copy unit 126 of FIG. 1, in accordance with one embodiment.
The synchronization unit 120 includes a write sample enable 140, a
write pointer compare unit 142, advance hold logic unit 144, an
write sample exclusive OR gate 146, a valid stage 1 unit 148, an
advance stage 2 unit 150, an advance stage 3 unit 152, an advance
stage 4 unit 154, and an advance edge exclusive OR gate 156. The
synchronization unit 120 also includes a valid stage 2 unit 170, a
valid stage 3 unit 172, a valid stage 4 unit 174, a valid edge
exclusive OR gate 176, and a write copy enable 178.
[0031] In operation, if the write pointer is not the same as the
write sample in the write sample unit 114, the write sample enable
140 turns on (e.g., outputs a "1"), which enables the write sample
unit 114 to sample the write pointer and causes the advance hold
logic 144 to toggle the value at valid stage 1 unit 148. This makes
the write pointer and the write sample equal. The write pointer
compare unit 142 outputs a "0," which causes the advance hold logic
144 to turn off the write sample enable 140, thus allowing the
write sample unit 114 and valid stage 1 unit 148 to hold their
values until an acknowledge is received and a new write pointer is
written. When the write pointer and write sample are different, the
advance hold logic 144 turns on write_sample_enable, which clocks
the write pointer into the write sample and toggles the value in
valid stage 1. The next cycle, when the write pointer and write
sample are equal, advance hold logic 144 turns off
write_sample_enable, which allows write sample and valid stage 1 to
hold their values. Write_sample_enable is designed to be a single
cycle pulse (only on during the cycle the pointer is copied from
the write pointer to the write sample). The valid stage 1 unit 148
is a register that stores values (i.e., "1" or "0"), and because of
the feedback loop through the write sample exclusive OR gate 146,
the valid stage 1 unit 148 holds its stored value until the valid
stage 1 is again toggled by the advance hold logic 144. In an
example scenario, if all of the advance and valid stage units
currently hold "0"s and the valid stage 1 unit 148 is toggled from
a "0" to a "1," the rest of the advance and valid stage units also
toggle to a "1" in the sequence shown in FIG. 3. In one embodiment,
a delay may be imposed before transmitting data to the valid stage
1 148 and to the rest of the advance and valid stage units. While
FIG. 3 shows a particular number of stage units (e.g., 7), the
specific number of stage units will depend on the number of stage
units required to resolve metastability in a particular
application. Every additional stage unit provides extra time for
data to stabilize, and the multiple stages of registers guarantees
that the signal is stable. By the time the 0-to-1 or 1-to-0
transition is seen at valid Stage 3, the signal is stable, thus
safe to bring across the multi-bit write pointer (then when the
destination sees the new write pointer value, the data in the
buffer has also been stable). The valid edge exclusive OR gate 176
functions as an edge detector that detects changes from a "0" to a
"1" or from a "1" to a "0." When a change is detected, the valid
edge exclusive OR gate 176 generates a single-cycle pulse that
turns the write copy enable 178 on. This enables the write copy
unit 126 to store the write sample from the write sample unit
114.
[0032] FIG. 4 is a block diagram of the advance hold logic unit 144
of FIG. 3 in accordance with one embodiment. The advance hold logic
unit 144 includes an advance hold unit 180, an inverter 182, AND
gates 184 and 188, and an OR gate 190. In one embodiment, the
source may receive an acknowledgement, also referred to as an
advance signal, from the destination. The advance signal
acknowledges that the sampled write pointer was copied to the write
copy unit 126. If the advance signal is received before the write
pointer changes again, the source holds the advance signal until
the pointer changes. The destination will know when the write
pointer changes again; because once the write pointer changes, it
is sampled and the destination is signaled that the source's
sampled write pointer has been updated. In some cases it is
desirable to delay this signal (that the write pointer has been
changed again) for some number of cycles after the write pointer
changes. This avoids signaling to the destination after only a
single entry has been written by the source to the buffer 102.
Accordingly, this delays the destination sending the advance signal
to the source. By holding/delaying the advance signal, the source
may write several entries in the buffer 102 before signaling the
destination that the write pointer has been changed again. This
helps reduce the effects of a "startup delay" of an empty buffer.
In one embodiment, the number of cycles the advance signal is held
in this situation is determined by a programmable register such as
the delay init unit 202 described below in connection with FIG.
5.
[0033] FIG. 5 is a block diagram of an advance hold logic unit 200,
which may be used to implement the advance hold logic unit 144 of
FIG. 3, in accordance with another embodiment. The advance hold
logic unit 200 of FIG. 5 is similar to that of FIG. 4, except that
the advance hold logic unit 200 of FIG. 5 also includes a delay
initialization unit 202, a multiplexer 204, a decrementer 206, a
comparator 208, an advance delay unit 210, and AND gates 212 and
214. The advance hold logic unit 200 also includes an advance hold
unit 180, an inverter 182, AND gates 184 and 188, and an OR gate
190.
[0034] In one embodiment, the size and depth of the buffer 102
(FIG. 1) may be optimized for particular ranges of frequencies such
that the interface does not limit data movement during steady state
reads and writes. Embodiments function properly, if not optimally,
regardless of frequency/phase relationships between the clocks. In
one embodiment, the buffer 102 may be optimized for a "nominal"
(typical) operating point (i.e., some frequency relationship
between the domains) such that the slower side (source/destination)
limits data movement. As a result, the buffer 102 would not be the
limiter.
[0035] In one embodiment, the read pointer on the destination side
can implement a hold delay on its advance signal coming from the
source, in a similar fashion described above for the write pointer.
This helps to reduce the effects of a "startup delay" when an
inactive destination suddenly wakes up and starts reading a full
buffer.
[0036] As indicated above, a single-bit handshake is used for
synchronization. In one embodiment, a single valid bit triggers a
sampling of the full pointer when it is known to be stable. This
allows any number reads/writes to occur on the faster side before
being sampled by the slower side. As a result, bursty read-write
behavior at high clock ratios (greater than 2:1) is not throttled
by deficiencies in the interface. This also ensures that the read
and write pointers do not change while they are sampled. The write
side of the interface may fill the buffer as long as a full
condition is not met and the read side may read data until an empty
condition occurs. This allows the source to continue filling the
buffer while waiting for an acknowledgement that the data was
received by the destination. The destination is able to read
multiple entries from the buffer while updating the source on the
status of the accepted data. This leads to a high bandwidth
interface.
[0037] FIG. 6 shows a block diagram of an exemplary design flow 600
used for example, in semiconductor design, manufacturing, and/or
test. Design flow 600 may vary depending on the type of IC being
designed. For example, a design flow 600 for building an
application specific IC (ASIC) may differ from a design flow 600
for designing a standard component. Design structure 620 is
preferably an input to a design process 610 and may come from an IP
provider, a core developer, or other design company or may be
generated by the operator of the design flow, or from other
sources. Design structure 620 comprises the circuits described
above and shown in FIGS. 1 and 3-5 in the form of schematics or
HDL, a hardware-description language (e.g., Verilog, VHDL, C,
etc.). Design structure 620 may be contained on one or more machine
readable medium. For example, design structure 620 may be a text
file or a graphical representation of a circuit as described above
and shown in FIGS. 1 and 3-5. Design process 610 preferably
synthesizes (or translates) the circuit described above and shown
in FIGS. 1 and 3-5 into a netlist 680, where netlist 680 is, for
example, a list of wires, transistors, logic gates, control
circuits, I/O, models, etc. that describes the connections to other
elements and circuits in an integrated circuit design and recorded
on at least one of machine readable medium. For example, the medium
may be a storage medium such as a CD, a compact flash, other flash
memory, or a hard-disk drive. The medium may also be a packet of
data to be sent via the Internet, or other networking suitable
means. The synthesis may be an iterative process in which netlist
680 is resynthesized one or more times depending on design
specifications and parameters for the circuit.
[0038] Design process 610 may include using a variety of inputs;
for example, inputs from library elements 630 which may house a set
of commonly used elements, circuits, and devices, including models,
layouts, and symbolic representations, for a given manufacturing
technology (e.g., different technology nodes, 32 nm, 45 nm, 90 nm,
etc.), design specifications 640, characterization data 650,
verification data 660, design rules 670, and test data files 685
(which may include test patterns and other testing information).
Design process 610 may further include, for example, standard
circuit design processes such as timing analysis, verification,
design rule checking, place and route operations, etc. One of
ordinary skill in the art of integrated circuit design can
appreciate the extent of possible electronic design automation
tools and applications used in design process 610 without deviating
from the scope and spirit of the invention. The design structure of
the invention is not limited to any specific design flow.
[0039] Design process 610 preferably translates a circuit as
described above and shown in FIGS. 1 and 3-5, along with any
additional integrated circuit design or data (if applicable), into
a second design structure 690. Design structure 690 resides on a
storage medium in a data format used for the exchange of layout
data of integrated circuits (e.g. information stored in a GDSII
(GDS2), GL1, OASIS, or any other suitable format for storing such
design structures). Design structure 690 may comprise information
such as, for example, test data files, design content files,
manufacturing data, layout parameters, wires, levels of metal,
vias, shapes, data for routing through the manufacturing line, and
any other data required by a semiconductor manufacturer to produce
a circuit as described above and shown in FIGS. 1 and 3-5. Design
structure (1290) may then proceed to a stage 695 where, for
example, design structure 690: proceeds to tape-out, is released to
manufacturing, is released to a mask house, is sent to another
design house, is sent back to the customer, etc.
[0040] According to the system and method disclosed herein, the
present invention provides numerous benefits. For example,
embodiments of the present invention enable data movement across
asynchronous clock domains without relying on fixed frequency or
phase relationships between the domains. The interface works
properly under any frequency and phase differences, and under
dynamically changing frequency and phase differences. Embodiments
of the present invention also sustain maximum throughput during
steady-state data movement.
[0041] A data interface system has been disclosed. In one
embodiment, the system includes a buffer that receives data from a
source in a source clock domain and stores the data to be read by a
destination in a destination clock domain. The buffer functions in
both the source clock domain and the destination clock domain. The
system also includes a write pointer that points to data written by
the source and includes a read pointer that points to data read by
the destination, where the write pointer and the read pointer
enable the data to be transmitted from the first clock domain to
the second clock domain asynchronously.
[0042] The present invention has been described in accordance with
the embodiments shown. One of ordinary skill in the art will
readily recognize that there could be variations to the
embodiments, and that any variations would be within the spirit and
scope of the present invention. Accordingly, many modifications may
be made by one of ordinary skill in the art without departing from
the spirit and scope of the appended claims.
* * * * *